The Great Proto-Indo-European Thread

[linguoboy]

Anyone know of a good source for information on PIE diminutive suffixes?

[Zju]

Bringing the issue of the vowel system back, has anyone considered <e o> being /ə a/? It could have gone like this:
1. [ə a h₂ə h₂a h₃ə h₃a]
ə changes to a next to h₃
2. [ə a h₂ə h₂a h₃a h₃a]
a becomes more lax, gradually shifting to o; ə next to h₂ shifts to a to fill the gap
3. [ə o h₂a h₂o h₃o h₃o]
ə changes to e to finish the transition to a more commen vowel system
4. [e o h₂a h₂o h₃o h₃o]
laryngeals dissappear
5. [e o a o o o]

There you have it, common vowel systems at both ends of the transition. This hypothesis only requires that h₃ had more influence on the vowels than h₂, maybe voiced vs. voiceless contrast.

[WeepingElf]

Bringing the issue of the vowel system back, has anyone considered <e o> being /ə a/? It could have gone like this:
1. [ə a h₂ə h₂a h₃ə h₃a]
ə changes to a next to h₃
2. [ə a h₂ə h₂a h₃a h₃a]
a becomes more lax, gradually shifting to o; ə next to h₂ shifts to a to fill the gap
3. [ə o h₂a h₂o h₃o h₃o]
ə changes to e to finish the transition to a more commen vowel system
4. [e o h₂a h₂o h₃o h₃o]
laryngeals dissappear
5. [e o a o o o]

There you have it, common vowel systems at both ends of the transition. This hypothesis only requires that h₃ had more influence on the vowels than h₂, maybe voiced vs. voiceless contrast.

This makes sense, and may have been what actually happened in the language. I see no problems with it right now.

[Particles the Greek]

How many of you who've put forward hypothesises here hope they'll be taken up by more serious PIE scolarship (i.e the kind people get paid for)? Because quite a few of them seem perfectly reasonable to me and deserve proper consideration.

[WeepingElf]

How many of you who've put forward hypothesises here hope they'll be taken up by more serious PIE scolarship (i.e the kind people get paid for)? Because quite a few of them seem perfectly reasonable to me and deserve proper consideration.

I frankly don't really expect this. Some of the ideas discussed here may make sense as genuine scholarship (though some perhaps overlook problems nobody here is aware of), but many professionals look down on amateurs in general, practically equating them with crackpots. In Indo-European studies, the opinion is especially popular that we can just not look any deeper than Late PIE, and people who concern themselves with internal reconstruction, or, worse, with comparisons between IE and other language families, are met with suspicion even if they have academic credentials. (Of course, academic credentials are not a sure sign that one is not dealing with a crackpot, either!)

[KathTheDragon]

What's the opinion here on the 3rd dorsal series of stops, the front velars? Real or not?

[Cedh]

The front velars are very much real. In fact, they're sort of the "default dorsals", being much more common than the plain velars and somewhat more common than the labiovelars. They might have been pronounced as true velars though, with the "plain velars" being realized as uvulars or something similar.

The actual question should be: What about the plain velars? If any of the dorsal series can be explained away through conditioned allophony, it would be the plain velars. They're much less common than the other two series, and they seem to appear in restricted environments only (cf. Lehmann 1952).

[Particles the Greek]

In general, what do people here think of Szemerényi's (often persuasive) opinions that the laryngeals are a delusion, and that PIE always had five distinct vowels, long syllabic resonants, and voiceless aspirated stops?

[KathTheDragon]

The actual question should be: What about the plain velars?

Yeah, I meant as contrasting with the 'plain' velar series. In other words, two or three series?

[WeepingElf]

In general, what do people here think of Szemerényi's (often persuasive) opinions that the laryngeals are a delusion, and that PIE always had five distinct vowels, long syllabic resonants, and voiceless aspirated stops?

Not much. Szemerényi just was a stubborn old man who fought the windmills of progress in his discipline.

[WeepingElf]

The actual question should be: What about the plain velars?

Yeah, I meant as contrasting with the 'plain' velar series. In other words, two or three series?

I think there is good evidence for three series, even if only Albanian and Luvian show distinct reflexes of all three, and all others have merged the "plain" velars with one of the other two series.

[CatDoom]

Even if most languages don't have reflexes of all three series, the satem/centum split seems like pretty strong evidence that there was some kind of distinction between k and ḱ. Speaking of which, I know that k and kʷ are supposed to have merged in Proto-Indo-Aryan; are there any other satem languages where traces of that distinction survived?

[KathTheDragon]

There is unquestionably a distinction, but was it phonemic in PIE? I've read that every Centum language merged the two in exactly the same way, which implies either it happened before they all split, which iirc is impossible, or they never had the distinction.

[CatDoom]

It was my impression that ḱ > k was basically the defining feature of 'proto-centum,' whereas 'proto-satem' had ḱ > some kind of palatalized fricative or affricate. There probably never were any such languages, but ḱ > k happening in every member of a closely-related group of dialects doesn't seem unreasonable. In any event, is there a competing proposal for what might have conditioned the satem shift?

[KathTheDragon]

From what I remember, ḱ > k happened in every member of a non-contiguous group, which are commonly grouped as centum languages. Their point that that is highly improbable, if not impossible, is contingent on the centum languages not being one blob, as I believe it was Toch. A or B being a centum language in the middle of the satem languages. They back themselves up by showing that there were multiple different methods of palatalisation in the satem languages, leading to multiple different results, which is what you'd expect for common innovations.

[WeepingElf]

This matter is something where the wave theory applies well. When a language breaks up, it first becomes a dialect continuum before the dialects develop into distinct languages. Such a dialect continuum may stretch across thousands of miles and last up to 2000 years if not longer. So, around 1500 BC, there still may have been a vast Indo-European dialect continuum stretching all the way from the Rhine to the Indus. While an Indo-European speaker from the Rhine and one from the Indus were not able to understand each other's speech, there was no point where the languages of neighbouring villages were mutually unintelligible.

In such a dialect continuum, innovations such as centumization or satemization can propagate from dialect to dialect, and this seems to be what happened. The centum-ness of Tocharian, however, is an oddity which is to be explained either with a migration from west to east, or (more likely) a second centre of centumization - a merger of /k/ and /q/ is not a particularly odd change.

[Chagen]

In general, what do people here think of Szemerényi's (often persuasive) opinions that the laryngeals are a delusion, and that PIE always had five distinct vowels, long syllabic resonants, and voiceless aspirated stops?

How could one possibly state that when the languages all show clear signs of laryngeals being there? They show up in Hittite and in the other languages resulted in the long vowels and in Sanskrit they resulted in /i/ popping up in places it normally shouldn't (some roots add -i before suffixing the past participle suffix -ta, for instance, there's some other instances).

That's pure crackpot stuff. The amount of evidence in favor of laryngeals is massive. Yes, we have no ACTUAL proof, but we have no concrete proof PIE existed either (such as writing), yet we don't need that to be certain it did exist. Occam's Razor at its finest.

[CatDoom]

Plus you've got the "triple reflex" in ancient Greek, where three different vowels (*e, *a, and *o) show up in various places where other languages have only *a or *i, as well as in pre-consonant positions where no other IE language besides Armenian has *any* vowel. And yes, you could try to attribute the latter fact to some kind of widespread vowel syncope, but you'd have to bend over backward pretty far to come up with some other conditioning factors for the loss of those vowels. The laryngeal theory is remarkably parsimonious; it provides the simplest known explanation for a *ton* of different sound changes.

[Basilius]

The centum-ness of Tocharian, however, is an oddity <...>

Also, Anatolian. No, Centum as a valid grouping (genetic or areal) doesn't hold water.

How could one possibly state that when the languages all show clear signs of laryngeals being there?

I'm not Szemerényi and I don't know his arguments, but there are problems IMNSHO with proving that all those clear signs reflect the same type of entities in the protolanguage.

[Terra]

A few questions:
1) Does PIE have any roots of the form sC(C)V(C)C that *aren't* from s-mobile?
2) Does Anatolian/Hittite show any effects of s-mobile?
3) Does Anatolian/Hittite show any of the usual vowel changes (h2e > a/h2a) caused by laryngeals?

[Basilius]

A few questions:

None of them is easy...

1) Does PIE have any roots of the form sC(C)V(C)C that *aren't* from s-mobile?

Well, OK, there are tons of sC-roots (including very reliable ones, like 'stand' or 'snow') that haven't any attested forms without the s- (except in languages that regularly simplify such clusters).

On the other hand, there are quite a few roots which have reflexes with/without *s- in just one branch and without/with *s- in a few others. Since we don't understand the nature of the phenomenon well, we can't be sure that lack of forms with/without *s- for certain roots isn't accidental. Like, suppose it was a morphological process with somewhat restricted productivity already in the protolanguage; it may be very difficult to point to factors affecting the applicability of the process to specific roots, with the material at hand.

2) Does Anatolian/Hittite show any effects of s-mobile?

I don't know the answer, so I queried the Starling databases:

Code: Select all

http://starling.rinet.ru/cgi-bin/response.cgi?root=config&morpho=0&basename=\data\ie\piet&first=1&off=&text_proto=*%28s&method_proto=substring&text_meaning=&method_meaning=substring&text_hitt=&method_hitt=substring&ne_hitt=on&text_tokh=&method_tokh=substring&text_ind=&method_ind=substring&text_avest=&method_avest=substring&text_iran=&method_iran=substring&text_arm=&method_arm=substring&text_greek=&method_greek=substring&text_slav=&method_slav=substring&text_balt=&method_balt=substring&text_germ=&method_germ=substring&text_lat=&method_lat=substring&text_ital=&method_ital=substring&text_celt=&method_celt=substring&text_alb=&method_alb=substring&text_rusmean=&method_rusmean=substring&text_refer=&method_refer=substring&text_comment=&method_comment=substring&text_any=&method_any=substring&sort=proto

It seems that there are more PIE words with s-mobile whose Hittite cognates have no s-, and very few reliable ones beginning in isC- in Hittite (like iskalla- (I/II) 'zerreißen, aufschlitzen, abreißen'). Besides, there are iskar- (II) 'stechen; hineinstecken' (Tischler 399ff); karš-, karšija- 'abschneiden' (517 ff), which, if you ask me, are from what looks like a cluster of contaminating PIE roots (thus, not a very convincing example of s-mobile in Hittite itself).

3) Does Anatolian/Hittite show any of the usual vowel changes (h2e > a/h2a) caused by laryngeals?

Be informed that this is a dangerous question: if you immerse in this material, you may quickly (1) find out that historical phonetics of Anatolian, essentially, does not exist, and even, you know, (2) get the impression that all the optimism about current laryngealist theories may be a bit precipitate.

For example, take the initial h- in Hittite. It seems that Hittite words with good PIE cognates include a lot which begin in ha- but very few (perhaps even none) which begin in he- (or hi- potentially reflecting an earlier *he-, if they're of late attestation).

Most of such words have PIE *a- in traditional reconstruction (which is the main source of the idea that Hittite h reflects the a-coloring laryngeal); a few have *o- (e. g. hastai- 'bone'); however, an initial (traditional) *e- doesn't seem to be forbidden for such roots. For example, the cognates of Hittite hara(n)- 'eagle', besides a few forms in *or- (and, interestingly, none unambiguously pointing to *ar-), include also Lithuanian erẽlis and the Celtic forms like Old Irish irar, Welsh eryr and Breton er.

(It may be also worth mentioning here that not all roots beginning in traditional PIE *a- have h- in Hittite, cf. the root in aku- c. 'stone', akuwant- 'stony'.)

The facts mostly look like the above; the rest is, you know, sectarian theories. As a conlanger, you are free to make whatever you wish out of this.

[Terra]

Well, OK, there are tons of sC-roots (including very reliable ones, like 'stand' or 'snow') that haven't any attested forms without the s- (except in languages that regularly simplify such clusters).

Very interesting. Hittite seems to have plenty of sC- words. Btw, I'm just wondering how far back these sC- roots go, and if in the end all sC- roots are caused by mobile-s, and pre-PIE had only a C(C)V(C)C root structure. Does anybody know what kind of root shapes Proto-Uralic had?

which, if you ask me, are from what looks like a cluster of contaminating PIE roots (thus, not a very convincing example of s-mobile in Hittite itself).

Yes, there seems to be contamination here. Imo, there were two separate roots 'sek' and 'ker' with similar meanings, and then s-mobile made 'sker' from the latter, causing more confusion and contamination.

And yeah, PIE has the remnants of what looks like suffixes that used to mean something, but then got fused into the root. (I asked about these earlier in the thread iirc.)

...

Another question about the root *mesg (meaning 'dip, sink, drown', yielded Latin 'mergo'). The 'sg/zg' part looks suspect. Could this be a compound of two other roots? Maybe like *meH+*seg ? I don't know any roots that would match the meaning though.

[Basilius]

Btw, I'm just wondering how far back these sC- roots go, and if in the end all sC- roots are caused by mobile-s, and pre-PIE had only a C(C)V(C)C root structure. Does anybody know what kind of root shapes Proto-Uralic had?

At any rate, PU had no initial clusters (and no prefixes).

Illich-Svitych's version of Nostratic treated PIE *sK- and *st- as mostly reflecting (two different series of) Nostratic affricates, but (IIRC) proposed no source for *sp- (which is found in a couple Indo-Hittite roots) or *sN-. Those affricates were supposed to yield affricate reflexes in PU.

The 'sg/zg' part looks suspect.

Why? There are a few other decent PIE roots with medial clusters of the shape s + a voiced obstruent.

[Terra]

Why? There are a few other decent PIE roots with medial clusters of the shape s + a voiced obstruent.

Well, I was thinking of 'nizd', which is 'ni' + 'sed'.

[Chagen]

Some questions:

Working on Proto-Pasuu has given me an interest in making an apost PIE conlang, but of course I need some books to do that. As good as Morrigan's fantastic font of ebooks are, I'll admit: I'm old-fashioned and like to hold dead tree stuff, especially when I'm working on conlangs as I do nearly all of that on paper.

With the suggestions I had received earlier on in this thread I'm definitely planning on picking up a copy of Indo-European Language and Culture by Benjamin W. Fortson IV. However, over this suggestion from Dewrad:

Arguably, for conlanging purposes, I would still say that the rather traditionalist "Brugmannian" reconstruction given in Szemerényi's Introduction to Indo-European Linguistics is more useful. It stops budding PIE-conlangers getting carried away by all the laryngeals, if nothing else.

Given that Wenetic is an awesome and high quality lang, anything suggested by Dewrad would probably be a good idea, but I'm still unsure how useful this book will be. Some internet searching shows that the book goes into really deep detail on PIE morphology, which is useful...but it also shows that Szemerényi was hell-bentedly opposed to many PIE conventions now taken for granted, such as laryngeals. As good as the book sounds...it's going to be quite odd without laryngeals, especially as for my hypothetical language I want to be a super-special snowflake and keep at least a few of the laryngeals; probably gonna have them turn into vowels when syllabic and next to a consonant (maybe with lots of wacky sound change a la Sanskrit's voiceless aspirates but far far worse), but intervocally, word-finally, and word-initially they'll stay as /h/ or /x/ or something.